Uploaded image for project: 'Network Observability'
  1. Network Observability
  2. NETOBSERV-1956

ConnTrack "All" performance is really poor

XMLWordPrintable

    • Quality / Stability / Reliability
    • False
    • Hide

      None

      Show
      None
    • None
    • None
    • None
    • None
    • NetObserv - Sprint 265
    • None
    • None
    • Hide
      Configuring FlowCollector with "processor.logTypes=All" consumes a lot more resources (CPU, memory and network bandwidth) than the other options. This was previously not documented. It is now documented, and will trigger a warning from the validation webhook.
      Show
      Configuring FlowCollector with "processor.logTypes=All" consumes a lot more resources (CPU, memory and network bandwidth) than the other options. This was previously not documented. It is now documented, and will trigger a warning from the validation webhook.

      Setting logTypes=All :

      cf https://issues.redhat.com/browse/NETOBSERV-1922?focusedId=25741876&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-25741876

      > FLP CPU up to +120% ("only" +70% in WIDE scenario), FLP memory around +40%, and similar figures for Loki, and storage +40%

      We should at least warn about that in the CRD doc, and perhaps also deprecate this mode.

              jtakvori Joel Takvorian
              jtakvori Joel Takvorian
              Amogh Rameshappa Devapura Amogh Rameshappa Devapura
              None
              Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

                Created:
                Updated:
                Resolved: